This work is a continuation of our pilot study on the use of off‐center s‐type Gaussian functions in noncovalent interaction calculations. A grid of s‐type Gaussians surrounding the molecule is intended to substitute for the presence of diffuse basis functions, which are important for accurate description of noncovalent interactions. The advantage over the use of diffuse functions is reduction or elimination of linear dependency issues in the atomic orbital basis set, which often causes convergence problems. Placement of a grid of s‐functions on the surface of the molecule allows more favorable scaling of the total number of basis functions with the molecular size to be achieved. In this article, we present several innovations on the concept, namely the parametrization and assessment of grids with triple‐ζ quality basis set; use in post‐second‐order correlation treatment (methods such as Møller–Plesset third‐order or coupled‐cluster); and combination with modified virtual orbital approaches, namely the frozen natural orbitals and optimized virtual orbital space methods.

Halogen bond (X-bond) is a noncovalent interaction between a halogen atom and an electron donor. It is often rationalized by a region of the positive electrostatic potential on the halogen atom, so-called σ-hole. The X-bond strength increases with the atomic number of the halogen involved; thus, for heavier halogens, relativistic effects become of concern. This poses a challenge for the quantum chemical description of X-bonded complexes. To quantify scalar relativistic effects (SREs) on the interaction energies and σ-hole properties, we have performed highly accurate coupled-cluster calculations at the complete basis set limit of several X-bonded complexes and their halogenated monomers. We found that the SREs are comparable in magnitude to the effect of the basis set. The nonrelativistic calculations typically underestimate the attraction by up to 5% or 23% for brominated and iodinated complexes, respectively. Counterintuitively, the electron densities at the bond critical points are larger for SRE-free calculations than for the relativistic ones. SREs yield smaller, flatter, and more positive σ-holes. Finally, we highlight the importance of diffuse functions in the basis sets and provide quantitative arguments for using basis sets with pseudopotentials as an affordable alternative to a more rigorous Douglas-Kroll-Hess relativistic theory.

No abstract